The biomechanical benefits of active sitting

Abstract This cross-sectional study examined the biomechanical effects of two active chairs (AC1: had the feature to pedal and slide forward on the seat pan; AC2: a multiaxial motion seat pan) compared to a traditional office chair and standing workstation. Twenty-four healthy participants worked at each of the workstations for 60-min. The following equipment was used: Motion Capture, Electromyography, Ratings of Perceived Discomfort Questionnaire, and Exit Survey. The active protocol had positive effects on the body, including increased neuromuscular activity in the gastrocnemius, increased overall movement, and a more open trunk–thigh angle. Greater discomfort in the buttocks due to the lack of seat pan contour was reported for the AC1 which identified a need for a design modification. While standing, participants’ shoulders were less flexed than when sitting in any of the three seats, however, greater discomfort was reported in the lower legs after 1 h of computer work. Practitioner summary: A comparison of four different workstations was conducted to further understand the use of active workstations. Active sitting was found to have positive effects on the body, such as allowing sitters to increase movement while sitting without the high activation of muscular activity. Standing can also provide a positive break from sitting.


Introduction
The vast majority of Canadians have a computer-based job, requiring workers to spend �71% of their 8-h shift in a static seated position (Clemes, O'Connell, and Edwardson 2014). Occupational sitting is the predominant cause of sedentary behaviour in many industrialised countries (Cole, Tully, and Cupples 2015). The increased health concerns associated with sedentary behaviour in the workplace have resulted in newly developed ergonomic tools to help promote movement. A current popular tool at the workplace involves the use of sit-to-stand desks, and now starting to arise is the use of active chairs. Recent designs of active chairs show promise in promoting movement at the workplace, however, empirical evidence is currently lacking.

Prolonged sitting
Prolonged sitting is defined as any period of continuous sitting for more than 30-min (Stranden 2000;Triglav et al. 2019), and has been linked to an increased risk of experiencing musculoskeletal discomfort (Schinkel-Ivy, Nairn, and Drake 2013). It is estimated that �80% of the Canadian population will experience at least one chronic low back occurrence in their lifetime (Rubin 2007), with 33-61.6% being seated workers (Spyropoulos et al. 2007). Sittingrelated low back pain (LBP) is predominantly associated with an adopted kyphotic or forward slouched/ leaning posture. A kyphotic posture occurs when the sitter's upper body is flexed forward (rounded shoulders) and no longer engages the back support, causing an malalignment of lumbar curvature (absent of the lordosis curve) (Makhsous et al. 2003). Back-related discomfort due to kyphosis has been linked to low-level muscle contraction in the lumbar erector spinae (McGill, Hughson, and Parks 2000;Callaghan and Dunk 2002;Beach, Mooney, and Callaghan 2003). A sustained 2% maximum voluntary contraction has been reported to impair the oxygen transport to the muscles, resulting in muscle pain and fatigue, this can also lead to involuntary micro-movement while seated on the chair (McGill, Hughson, and Parks 2000). Low levels of muscular activity also suggest that the sitter is relying on their passive tissues (ligaments, disc, fascia) surrounding the spine (Beach et al. 2005). According to Callaghan and Dunk (2002), the potential source of low back pain during seated work could be the reliance on the ligaments (which contain a large number of free nerve endings) while seated in a kyphotic posture. If the ligaments must sustain the load of the upper back for prolonged periods of time, due to a kyphotic posture, pain receptors could be activated due to viscoelastic creep applied on the posterior ligaments of the spine, posterior aspect of the intervertebral discs and facet capsules (Solomonow 2012;De Carvalho et al. 2010). Extended periods of sitting in a kyphotic posture have also been shown to increase perceived discomfort in the upper limbs (Jia and Nussbaum 2018).

Prolonged standing
Standing has become a recommendation as a resting approach from prolonged sitting. Some studies have found positive benefits associated with increasing standing time, such as better posture by promoting less trunk flexion when standing (Reiff, Marlatt, and Dengel 2012). However, standing for an extended period of time is not recommended; where prolonged standing is defined as more than 1 h of continuous standing coupled with 4 h of standing per work day, which is half of a work shift (Halim and Rahman Omar 2012;Tomei et al. 1999). Standing for an extended period of time during the day may cause the onset of low back pain (LBP) or discomfort (Tissot, Messing, and Stock 2009). Static standing places a compression load on the spine and with time triggers the muscle fatigue mechanisms surrounding the spine (Callaghan and McGill 2001), increasing the risk of the onset of LBP (Hasegawa et al. 2018). Fewster et al. (2019) observed how a group of individuals with LBP and a group experiencing no LBP reacted to a prolonged standing task. While participants were standing, they were asked to elevate one leg on a 13 cm foot rest for 1 min, every 3-min, for 80-min. Fewster et al. (2019) concluded that changing position with a 1:3 ratio, helped mitigate LBP by increasing lumbar flexion and increasing glutaeus medius co-activity when using a footrest, compared to a level standing. The group with no LBP experienced no pain over time and individuals with LBP did experienced significantly less muscle fatigue and low back discomfort towards the end of the trial (Fewster et al. 2019). Although alternating positions while standing reduced perceived low back discomfort, Bailey and Locke (2015) found that standing alone is not a great rest from sitting when compared to a walking break. Bailey and Locke (2015) found that over a 4-h sitting period, a 2-min walking break is better than a 2-min standing break, every 30 min because light physical activity will help lower glycaemia (blood sugar) levels amongst healthy adults.

Dynamic/active sitting
The definition of dynamic and active sitting has changed over the past few years. Dynamic chairs are defined as where the sitter is in constant motion (Cardoso, Cardenas, and Albert 2021) like a rocking chair (Pynt 2015). When comparing a dynamic chair to a traditional office chair, there is no significant difference in trunk kinematics or trunk extensor musculature (van Die € En, De Looze, and Hermans 2001) and no significant different in physical activity and posture (Ellegast et al. 2012). On the other hand, active chairs are designed so that the sitter is providing the action to move the chair, using alternating muscle groups, while the mechanism of the chair accommodates that action (Cardoso, Cardenas, and Albert 2021;Pynt 2015). The purpose of an active chair is to reduce static muscle activity and perceived discomfort.
To the author's knowledge, three different active chairs have been investigated in published research: 1-a multiaxial chair that simulates a stability ball with a low back support (Koepp, Moore, and Levine 2016;Triglav et al. 2019), 2-a saddle chair (Gadge and Innes 2007;Barrett 2019), and 3-a traditional office chair with a modified split seat pan (Cardoso, Cardenas, and Albert 2021;Cardenas et al. 2022): The multiaxial chair simulates a stability ball by allowing seat pan rotation of 14 � in all directions. The multiaxial chair has a low back support and no arm rests. The multiaxial chair has been found to reduce spine flexion (Barrett 2019). Working in a posture that helps preserve the natural lordotic curve, has been found to help reduce transient LBP (Gadge and Innes 2007), and strengthen the core muscles (Pynt 2015). Barrett (2019) compared the multiaxial chair to a traditional office chair and concluded that individuals had less flexion in the spine while seated, perceived less overall discomfort, and had more seat pan movement on the multiaxial chair. Muscle activity and peak sitting pressure were similar for both chairs.
A saddle chair was created to improve posture and reduce LBP (Gadge and Innes 2007). As the name of the chair implies, this design is a stool with the seat pan shaped like a saddle (with or without a low back rest). Currently, on the market, some saddle chairs are stationary and others are active and vary in the degree of stability when sitting. Both static and active designs require less back muscle activation compared to a traditional chair while seated. Lower levels of muscle activation enable the sitter to tilt the pelvis anteriorly and have a more neutral curve in the cervical and lumbar region of the spine (Annetts et al. 2012;O'Sullivan et al. 2012). However, over time, due to the variance of pressure distribution associated with the saddle chair seat pan, blood circulation was reduced, creating discomfort in the lower limbs, hip, and buttocks area (Gadge and Innes 2007;Synnott et al. 2017).
The split seat pan design has a longitudinal split seat pan allowing the sitter to simultaneously alternate their feet between plantar/dorsi flexion and hip flexion/extension to simulate the action of walking and has been involved in two separate studies, in the same research laboratory: Cardoso, Cardenas, and Albert (2021) and Cardenas et al. (submitted).
In the first study (Phase 1), two active chairs (AC1 and AC2) and a traditional office chair were analysed (Cardoso, Cardenas, and Albert 2021). The AC1 chair had a split seat pan design and participants were able to move their feet in plantar and dorsiflexion, essentially peddling their feet while seated. The AC2 was a modified version of the AC1, with a pivot swivel at the front end of the seat pan to encourage side-to-side swaying while seated (both chairs were different than this current study). To gather a true representation of how individuals used active chairs in an office setting, Cardoso, Cardenas, and Albert's (2021) study allowed participants to choose if/when they pedal their feet which most likely contributed to a lack of significant differences found in their study.
The second study (Cardenas et al. 2022), evaluated a new and updated design of the AC1 active chair (shorter seat pan and a more defined contour) and compared the active chair to a standard office chair and a standing desk (this AC was also different than this current study). Because the changes observed using the AC1 chair in Phase 1 was minimal, a standardised protocol was developed in Phase 2 to evaluate the active chair's maximum potential. In Phase 2 participants were required to pedal their feet at a specified tempo of 40 beats per minute while seated in the AC1 and following the rhythm of a metronome. The greatest change found in this study was the physiological alterations; specifically, greater changes in oxygenated blood was found in the calf muscles (gastrocnemius) compared to Phase 1.
The purpose of the study was to compare sitting biomechanics between active seating, traditional office seats, and standing workstations. This research study had the following research hypotheses: H1-Acknowledging that both active seats have different motion patterns, it is hypothesised that the active chairs will promote greater whole-body movement patterns compared to a traditional office chair. H2-The active chairs will require higher muscle activity in the lower extremities and abdominal muscles. H3-The active chairs will be led to less perceived wholebody discomfort than either a traditional office chair or standing. The authors would also like to acknowledge that this manuscript is derived from L� eger's (2021) master's thesis.

Participants
Twenty-four healthy participants were recruited from a university student population: twelve males (age: 22.92 ± 3.00 years, height: 176.73 ± 5.77 cm, weight: 78.62 ± 6.66 kg), and 12 females (age: 21.08 ± 0.79 years, height: 161.03 ± 4.07 cm, weight: 61.21 ± 7.35 kg), between 19 and 40 years of age. A G-power analysis was performed to determine the sample size of twenty-four participants. To be eligible to participate, individuals had to fill out a health screening questionnaire, with an outcome of 3-month history of no pain or injury in the low back region; no vascular disorders in the lower limbs; and the ability to stand for 60-min or longer. All 24 participants had no past experience with active chairs, however, it was not a requirement. Before participation, participants were provided a detailed letter describing the study and they had an opportunity to ask questions before providing written consent for the research ethics board (REB) approved project. This project has been reviewed by the Research Ethics board of the University of New Brunswick and Universit� e de Moncton.

Workstation comparisons
Four different workstations ( Figure 1) were compared: 1-an active chair with a split seat pan design (Active Chair 1: AC1); 2-an active chair with a multiaxial design (Active Chair 2: AC2); 3-a traditional office chair (Control) and a standing workstation (Desk). The new active chair (AC1) had a forward-reclining mechanism that enabled the sitter to open the hip-thigh angle by sliding the pelvis forward while the head and shoulders remain stationary. The seat pan had a split longitudinally, which was designed to promote lumbar, pelvic, and hip motion by alternating active ankle plantar/dorsiflexion and a 10 � range of alternating hip flexion/extension similar to a walking action. The AC1 was engineered to increase motion, open the trunk-thigh angle and reduce compressive loading of the spine. The height of the AC1 had a ranged from 18 inches to 20.75 inches; a seat pan depth of 17 inches, and width of 21 inches and backrest height of 22 inches, and width of 20.5 inches. The active multiaxial chair (AC2) had a low back support (height: 8 inches, width: 15.5 inches) and was designed to simulate an exercise ball. The multiaxial chair encouraged the movement of 14 � in all directions, providing postural support, and had an indentation in the seat pan as a means to offload the sacrum. For the study, two models were used, a small model for individuals who were 5 0 5 00 (or shorter) and a large model for individuals who are 5 0 6 00 (or taller). The height of the AC2 had a range from 17.1 to 19.2 inches for the small model and 18.7 to 22.3 inches for the large model; both models had a seat pan depth range of 16.7-18.3 inches and a seat pan width of 19.3 inches and the backrest seat pan depth range of 16.7-18.3 inches, a seat pan width of 19.3 inches and the backrest. The traditional office chair (Control) had no active component, the seat pan height was adjustable, the back rest was reclined to 10 � (Wang et al. 2019) and the armrests were removed. The seat pan was set to a positive inclined angle of 5 � from the horizontal plane (Wang et al. 2019). The height of the Control chair had a range from 17 to 19.5 inches; a seat pan depth of 19 inches, and width of 20.5 inches and backrest height of 22.5 inches, and width of 19 inches. Lastly, the standing desk (Desk) was freely adjustable from 22 00 to 48 00 . The desk was 63 00 wide x 31 1 = 2 00 inches deep and was designed to fit the 5th percentile female to the 95th percentile male.

Experimental setup (Figure 2)
Participants were asked to visit the laboratory on two separate occasions with a minimum of 72 h of rest between each day. To help mitigate the development  of fatigue, two of the four workstations were investigated on each collection day. Consent forms and measurements of the participant's anthropometry were taken on the first session only, which included: weight, waist circumference, and height with and without shoes. Each participant had the office chair and desk adjusted for their anthropometrics using the Canadian Centre for Occupational Health and Safety (CCOHS) ergonomic guidelines (Canadian Centre for Occupational Health and Safety 2018). Participants were given the choice to mouse with their preferred handedness; although not all participants were righthanded, all 24 participants selected the right hand while they moused.

Electromyography (EMG)
The Bortec Octopus AMT-8 (Bortec Biomedical Inc, Calgary AB, Canada) was used to measure neuromuscular activity with the recording frequency set at 1024 Hz, during all four conditions. The myoelectric signal was collected in 4 � 15-min increments for a total 60-min. Eight unilateral muscles ( [G]. A ground electrode was placed on the right collarbone. Electrode placement was verified with palpation and instruction from Surface Electromyography for the Non-invasive Assessment of Muscles project guidelines (Seniam guidelines were followed) (Hermens et al. 2000). All bipolar electrodes were pre-gelled, disposable, and placed over the muscle's belly. The electrodes were 10 mm in diameter with an inter-electrode spacing of 19 ± 1 mm (centre to centre). To insure uniformity of spacing, alligator clips were used to help facilitate a secure connection with the electrode lead.
In both experimental sessions, maximum voluntary contractions (MVC) were administered to provide normalisation of electromyography results. Six isometric maximal voluntary contractions were performed by the participants. For each MVC, the participant held an isometric contraction for 5 s. Another MVC repetition was performed when the participant felt that they did not achieve maximal effort during the first attempt (Callaghan and McGill 2001;Gregory et al. 2006). When the MVC had to be repeated, a 2-min rest between each exertion was given. MVCs for the neck (SC) required participants to extend their neck against resistance while their torso was secured to the table (Cardoso, Cardenas, and Albert 2021). The resistance to the neck consisted of a metal chain attached to the ground and secured around the head with a strap. MVCs for the thoracolumbar extensors (TES, LES) included a maximal back extension (Cardoso, Cardenas, and Albert 2021). The participant's torso was suspended off the end of the table with their anterior-superior iliac spine (ASIS) on the superior edge of the table. The participants were required to perform thoracolumbar extension against manual resistance applied by one of the researchers (Cardoso, Cardenas, and Albert 2021). MVCs of the trunk flexors (EO) required a series of forward-flexion contractions coupled with a left trunk twisting with flexion (Cardoso, Cardenas, and Albert 2021). The participants were seated on a table with their knees flexed to �90 � and the trunk to �45 � from the horizontal. Manual resistance was provided by one of the researchers restricting flexion and twisting movement, while another researcher secured both feet to the table. MVCs of the rectus femoris (RF) and vastus lateralis (VL) required the participants to be sitting with 90 � hip flexion and 90 � knee flexion (Akima and Ando 2017). For the MVC of the vastus medialis (VM), the knee angle was at 60 � (Coburn et al. 2005). Both MVCs for the RF, VL, and the VM, a strap was secured to the chair, and placed to the distal leg, just above the ankle providing resistance during knee extension. MVCs of the gastrocnemius required the participant to lay down with the knees extended. A rolled towel was placed between the table and the tibia and a rigid strap was placed around the sole of the foot and secured to a plinth, providing resistance while the participant performed a plantar flexion (Reid et al. 2012).

Motion capture system
Kinematic data was collected using a Qualisys Miqus (Qualysis, G€ otenborg, Sweden) motion-capture system equipped with eight 3-dimensional cameras to quantify: head, shoulders, hips, pelvis, knees, trunk-pelvis, and trunk-thigh angle. A total of 71 retro-reflective markers were attached with 3M double-sided tape on the participant's skin. Each bony landmark had a retro-reflective marker, totalling 40 markers, and eight reference clusters (31 markers), were placed on the following: upper arms, left hip, thighs, shank, and trunk. The reference clusters were square plates made of plastic with four retro-reflective markers (except for the hip cluster, it only had 3 markers) glued in each corner. The cluster plates helped reduce skin movement and were attached with 3 M double-sided tape on the participant's skin. Markers on the back of the head were attached using flexible medical tape on the back of the head, specifically on the occipital ridge. Marker placements were based on the C-Motion Wiki Documentation site (van Sint Jan 2007). The Qualisys motion-capture system recorded at a frequency rate of 200 frames-per-second (fps) in 4 � 15min increments, for a total 60-min.
Once the participants were outfitted with markers (this applies to both sessions), participants were instructed to stand in an anatomical position for 5 s before the first trial of the session to identify each marker on the bony landmarks to create a 3-D model using Visual 3D software. The participants were then asked to stand in a T position (arms in abduction) for 5 s and then they rotated each articulation in the following order: right foot, right knee, right hip, left foot, left knee, left hip, pelvis, both wrists, elbows, shoulders, and the head. The dynamic calibration was then used to create to a 3-dimensional model, this model helped identify the label of each marker within the QTM software.
Before the collection trial, 21 markers were eliminated to allow for the following: 1-the participants to interact with the backrest without markers on the spine, therefore being able to fully engage with the seat, 2-to promote wrist movement while computing, and 3-to conduct the active motions of the chair without the hip and medial markers of the legs interfering with their movements. The following markers were removed: four markers on the spine (TV8, LV1, LV2, LV3), four markers of the pelvis (LIPS, RIPS, RIAS, LIAS), the markers on the xiphoid process (SXS), the forearm markers (RLA, LLA), both wrist markers (RRSP, RUSP and LRSP, LUSP) and all of the medial bony landmark (RHME, LHME, RFME, RTAM, LFME, LTAM). The 21 markers that were removed were virtually tracked in relation to the clusters and the local coordinate system.

Discomfort questionnaires
A Rate of Perceived Discomfort (RPD) questionnaire was used to monitor the participants perceived discomfort while computing. The RPD is a self-reported questionnaire with a 100-point visual analogue scale that indicates the level of discomfort relative to specific body parts (Cardenas et al. 2022;Cardoso et al. 2018, Cardoso, Cardenas, andAlbert 2021). A scale of point 0 corresponds to no discomfort and point 100 represents the worst discomfort imaginable; change from the baseline was calculated. Thirteen different body parts were measured with the RPD questionnaire on both the right and left sides of the body: shoulders, upper back, lower back, buttocks, upper and lower legs, as well as the unilateral measurement of the neck.

End of session survey
Following each trial, a survey of eight questions was administered to each participant to quantify their opinion of the workstation. The questionnaire was a 100-point scale with 0 corresponding to Not at all (positive response) and 100 corresponding to Extremely (negative response) (Menezes and Xavier 2018). One of the eight questions had a biomechanical focus and the results are presented in this communication. The questions were: 1-How affected by physical symptoms (pain, soreness, etc.) have I been in the last hour? Following each question, the participants had the opportunity to provide a brief explanation.

Protocol
Assessment of each workstation was randomised and the participants were asked to work in 15-min intervals for a total of 60-min, in each experimental trial. Before starting the collection trial, a rate of perceived discomfort questionnaire (RPD) was given as baseline.
After the second and fourth 15-min increments, an RPD questionnaire was administered, and at the end of the last 15-min increment an End of Session Survey was also administered. The participants were asked to do two different tasks during the trial: A 10-min typing task and a 5-min web browsing task. For each 15-min increment, participants began with the typing task and then proceeded to the web browsing task during the last 5-min; this format was repeated for each 15min increment. The active protocol was to alternate between a pedalling/Side-to-Side motion and sliding forward/Front-to-Back motion to the sound of a metronome operating at 40 bpm (Cardenas et al. 2022) at each minute during the web browsing task. The participants were instructed to always have at least one hand on the mouse or keyboard, and have both feet on the ground when seated. The standing protocol was similar to the sitting protocol where the participants were asked to always have at least one hand on the mouse or keyboard at all times when standing. Both feet had to remain in contact on the standing mats. These restrictions were implemented as a means to avoid a movie-watching position, such as having arms crossed with no interaction with the computer. Figure 3 provides a timeline of both experimental collections.

Electromyography
The raw signal was rectified (RMS converted) and Butterworth band pass filtered from 20-500 Hz to remove unwanted frequencies. Peak activity was found for each muscle during the MVC trials and used to normalise all subsequent EMG data. The average EMG data was then compiled into 1-min intervals to determine the level of muscle activity percentage change (from the MVC trials) during the 60-min of computing.

Motion capture system
Marker coordinates were gap filled and smoothed using a moving average of 5 frames/second. In Visual 3D (Germantown, MD, USA), the kinematic and joint angle data was filtered using a Butterworth low pass filter with a cut-off frequency of 6 Hz. The joint angles are illustrated in Figures 4-6 and were calculated in Visual 3D as follows: 1-The head angle was calculated using the head relative to the torso; 2-the shoulder angles represented the arms relative to the torso; 3-pelvis angle was relative to the local coordinate system; 4-the trunk-pelvis angle was represented by the trunk relative to the pelvis; 5-the hip angles were represented as the pelvis relative to the thigh; 6-the knees represented the anterior angle of the thigh relative to the shank; and 7-the trunk-thigh angle was represented by the posterior angle of the left thigh relative to the torso. The 7 joint angles followed an XYZ Euler rotation sequence to calculate flexion/extension, abduction/adduction, and axial rotation.
With the active protocol only during the web browsing task, joint angles were analysed differently to reflect a better understanding of posture while typing and movement while web browsing: 1. For the typing task (first 10-min), each joint angle data was normalised to 101 points and then subdivided into 10% cycles (representing 1-min), up to a 100% cycle (i.e. 10-min). During each 10% cycle, the average joint angles were determined for the following: head, shoulders, pelvis, trunk-pelvis, hips, trunk-thigh, and knees. 2. For the web browsing task (last 5-min), each joint angle data was compiled into five, 1-min intervals for the web browsing task and then normalised to a 101 points, for each 1-min interval. For each 1-min increment, the average of 10 specific range points (Max-Min points; one range point represented a full cycle of peddling) were calculated, giving us the average range change for each 1-min increment. The average range was calculated for the following joint angles: head, shoulders, pelvis, trunk-pelvis, trunk-thigh, hips, and knees (see Figures 4-6). The last 5-min were subdivided into 1 min intervals due to the different motions that were imposed during the active tasks.

Discomfort and end of session survey
The rate of perceived discomfort (RPD) questionnaire was administered before each trial and was set as the RPD baseline. The score change difference between RDP baseline and the RPD during the trial was compared (POST-PRE). The end of session survey was compared between each workstation.

Statistical analysis
A repeated measure mixed analysis of variance (ANOVA) with two within factors (independent variables: time and workstations) was performed to evaluate differences between movement patterns, neuromuscular activity, and perceived discomfort  (dependent variables). Before running the repeated measures ANOVA, each data point was plotted to observe if it was normally distributed. The Mauchly's test was used to assess the sphericity assumption. When the sphericity assumption was violated, the Greenhouse Geiser was used as a correction. The Alpha level was set at p < .05, and if any significant interactions were found, the Tukey correction was used as a post-hoc analysis.

Results
To better orient the reader, the following approach has been taken to present the results. The data were divided by task for the purposes of this analysis. The following nomenclature was used T#_##. The T symbolises the time point, followed by either a 15, the first 15-min increment, or 60, the fourth 15-min increment, and the number following the underscore represents the specific time point 10, 11, 12, 13, 14, or 15 min. For example, the 10-min typing task of the first and fourth increment, is identified as T15_10 and T60_10, respectively. For the 10-min typing task, EMG data were compiled and averaged into one point, and for the joint angles, into 10 points. For the 5-min web browsing task, the data was divided into five 1-min intervals due to performing two different motions on both active chairs. For the first increment, the data collected during the web browsing task is identified as T15_11, T15_12, T15_13, T15_14, and T15_15. For the last increment, the data collected during the web browsing task is identified as T60_11, T60_12, T60_13, T60_14, and T60_15. Each term represents the average data for 1 min.

Electromyography
Figures 7-9 provide an overview of the average percentage amplitude change from the maximal voluntary contraction at T60. In the attached link provided below, Supplementary Tables 1.1-1.8 provides the means and standard deviation (SD) of the % change scores and identified the determined significant differences between T15 and T60. Key statistical findings are bolted. Overall, participants required significantly more neuromuscular activity of the External Obliques using the Desk than the AC1, AC2, and the Control Chair  EMG findings for each of the four workstations (AC1, AC2, Control, and Desk) for each of the following unilateral muscles at T60_10: splenius capitis, erector spinae T9, erector spinae L3, external obliques, rectus femoris, vastus lateralis, vastus medialis, and gastrocnemius. The increase in EMG activity is a sign that more neuromuscular activity was being required. . EMG findings for each of the four workstations (AC1, AC2, Control, and Desk) for each of the following unilateral muscles at T60_14: splenius capitis, erector spinae T9, erector spinae L3, external obliques, rectus femoris, vastus lateralis, vastus medialis, and gastrocnemius. The increase in EMG activity is a sign that more neuromuscular activity was being required. at T15 and T60. When participants performed the Side-to-Side and Front-to-Back motion in both active chairs, AC1 and AC2, the gastrocnemius demonstrated significantly more neuromuscular activity than the Control Chair (Supplementary Table 1).

Motion capture system-joint angles
In the attached link provided below, Supplementary Tables 2.1-4.14 highlights the means and SD of the average angles for the typing task and the average range of motion for the web browsing task for both T15 and T60 of the X, Y, and Z Axis. Key statistical findings are bolted.
The following summarises our common findings at T15 and T60: For the typing task (first 10-min), The Desk required more head flexion than the AC2 and less shoulder flexion than the other three workstations. The AC1, AC2, and the Control Chair had a greater posterior tilt in the pelvis compared to standing. No significant difference was found between the three chairs for the hips, knees, and trunk-thigh angle in the X axis, and no significance was found in the Y (lateral bend) and Z (twist) axis for any joint angles. For the web browsing task (last 5-min), the AC1 provided a greater flexion/extension range of motion (ROM) in the trunk-thigh angle than the other three workstations when performing both motions, Front-to-Back and Side-to-Side.
The AC2 provided more flexion movement in the head and shoulders in Front-to-Back motion compared to the other three workstations, and more lateral bend and twist during the Side-to-Side motion. The AC1 and AC2 chairs provided more flexion/extension, lateral bend, and twist movement in the hips and knees compared to the Control Chair and The Desk when participants performed both movements, Front-to-Back and Side-to-Side.
When seated in an active chair, greater ROM in flexion/extension, lateral bend, and twist were found in the pelvis compared to the control chair. During Front-to-Back movement, AC2 promoted greater ROM in the pelvis than AC1 in flexion/extension. During Side-to-Side movement, AC2 promoted greater ROM change in the pelvis than AC1 in lateral bend.
Greater trunk-pelvis ROM was found in flexion/ extension using both active chairs vs. the Control Chair while moving Front-to-Back. During Side-to-Side movements, greater lateral movement and twist were found using both active chairs vs. the Control Chair. Figure 9. EMG findings for each of the four workstations (AC1, AC2, Control, and Desk) for each of the following unilateral muscles at T60_15: splenius capitis, erector spinae T9, erector spinae L3, external obliques, rectus femoris, vastus lateralis, vastus medialis, and gastrocnemius. The increase in EMG activity is a sign that more neuromuscular activity was being required.
Also found during Side-to-Side movements, AC2 also had greater lateral movement and twist compared to The Desk. During Side-to-Side movement, AC2 had greater trunk-pelvis ROM than the AC1.
Note: Due to the standing position being different than sitting, the hip, knee, and trunk-thigh angles were not compared to the three chairs (Supplementary Table 2). Figure 10 provides the rate of perceived discomfort scores at T60. Participants reported significantly higher discomfort scores in the left and right buttocks using the AC1 (

End of session survey
Participants ranked the Control Chair (10.04 ± 14.65) as less affected by physical symptoms after an hour of computing compared to the AC1 (22.21 ± 17.14, p ¼ .007) and Desk (37.33 ± 31.35 p < .001).

Discussion
The purpose of the current study was to compare the biomechanical aspects of two active chairs: a newly modified AC1 (split seat pan which allowed a pedalling and a sliding forward motion) and AC2 (a multiaxial chair, which simulated a stability ball, and used two specific motions, Side-to-Side and Front-to-Back) to a Control Chair and a standing desk (The Desk).

Neuromuscular activity
In general, active chairs (i.e. the AC1 and AC2 from the current study, and the saddle chair) are designed to free the sitter from the constrained postures of traditional upright office seating. Research on active chairs indicates positive results for increased leg muscle activation (Grooten et al. 2013) and improving the sitting posture (O'Sullivan et al. 2012). Sitting in an active chair can improve the upper body posture by constantly moving the pelvis, which results in rotating the spine in and out of a lordotic curve (Harrison et al. 1999). Micro-movement of the trunk muscles allows the sitter to contract and relax the muscles surrounding the spine, which increase disc nutrition (Holm et al. 1981) and decreases stress load on the soft tissue (Panjabi 1992) resulting in potentially decreasing muscle fatigue, and perceived discomfort in the low back (van Die € En, De Looze, and Hermans 2001). The constant movement makes it sustainable to maintain a proper neutral lordosis curve in the spine (Watanabe et al. 2014).
For this study, the two active chairs were designed differently. The AC1 was designed to engage the lower body muscles and the AC2 was designed to engage the core muscles. Neuromuscular data were collected on eight unilateral muscles, all on the righthand side of the participant. It was found that when the participants were seated in the AC2, they generally required significantly more neuromuscular activity in the SC and T9 when performing both Side-to-Side and Front-to-Back motions compared to the control chair. When standing, participants contracted their external obliques significantly more compared to the sitting. Standing also required the participants to contract their external obliques up to 12% of their maximal voluntary contraction, which is �4% higher than previously found by Cardenas et al. (2022). Potential reasons for the higher activity in the external obliques may be related to the cold temperatures of the laboratory environment, interference or the participants were also wearing minimal clothing, resulting in an increase in abdominal muscle contraction. No significant differences were found in the thigh muscles activity (Rectus Femoris, Vastus Lateralis, and Vastus Medialis) between all four workstations. On the contrary, Gao et al. (2017) found a large increase in neuromuscular activity in the thigh muscle when standing compared to sitting when computing for 2 h. Similarly to the current study's findings, Kuster, Bauer, and Baumgartner (2020) also found no significant difference in leg neuromuscular activities when comparing active chairs to a traditional chair. Kuster, Bauer, and Baumgartner (2020) analysed the effects of an active chair, with no backrest, that promoted Side-to-Side movement; the active chair also had the option to fix the seat pan in a static position. The experimental protocol in Kuster, Bauer, and Baumgartner (2020) evaluated three different conditions: 1-each participant sat in the active chair while maintaining a static neutral position; 2-participants had to perform six full cycles of Side-to-Side motion with a spontaneous range of motion, meaning the participant had the freedom to choose the extent of the range of motion; 3-participants were asked to perform another six cycles of Side-to-Side movement with the maximal range of motion. Kuster, Bauer, and Baumgartner (2020) reported that when participants used a spontaneous range of motion while computing, there were no significant differences in neuromuscular activity of the legs (Vastus Lateralis and Vastus Medialis) between the active chair and a static position. However, when the participants were performing the maximal range of motion on the active chair, there was a slight variation in the thigh muscles compared to the static position, due to contracting the legs and trunk muscles to support an upright position (to compute) (Kuster, Bauer, and Baumgartner 2020). Similarly to the current study, no significant differences were found in the upper leg neuromuscular activity between both active chairs, the control chair, and standing. Participants of the current study, potentially used a spontaneous range of motion when executing movement on the active chair like in Kuster et al. (2018) study, which would explain the similar results between both studies.

Joint angles
Joint angles were analysed differently for each task. For the 10-min of the typing task, each joint angle was represented as the average over each minute of typing. The following description represents the common findings at both T15_10 and T60_10 (typing task). For the head angle, standing required more neck flexion/extension than the AC2 at T15 and T60. Previous research found that a standing position added a greater gravitational demand on the head and neck when computing, resulting in a greater head flexion position (Ailneni et al. 2019;Shaghayegh Fard et al. 2016). For the trunk-thigh angle, hip and knee angles, no significance was found between the AC1, AC2, and the control chair, which indicated that participants were not moving on either of the active chairs when given the choice to actively move or not. For the shoulders, holding the mouse with the right hand required an additional 10 � of flexion compared to the left side. Since participants were closer to the table when computing than when seated, standing required 10 � less shoulder flexion than all three chairs. For the pelvis and trunk-pelvis angle, a significant difference was found for the desk, where the seated position required a greater posterior pelvis angle than standing. As seen in previous literature, a seated position increases the posterior pelvic tilt (Carter and Banister 1994) compared to a more anterior tilt when standing (De Carvalho et al. 2010). A less posterior pelvic tilt is considered to be the ideal seated position of the pelvis, however, to maintain the ideal pelvic position, it requires a low-level trunk muscle contraction, which results in muscle fatigue over time (Carter and Banister 1994). When muscle fatigue occurs, discomfort levels start to rise and provoke the sitter to change position into a slump or kyphotic relaxed posture, where the pelvis tilts posteriorly, which decreases the lordosis curve and reduces the activation of the spinal stabilising muscles (O'Sullivan et al. 2002).
For the 5-min web browsing task, the average range of motion in all 3 axes for each minute was analysed. The following description represents the common findings at both T15 and T60 for the web browsing task. No significance was found in head flexion/extension motion. When participants were performing the Sideto-Side motion on the AC2 chair, significance was found due the whole-body moving from Left-to-Right, the head was required to move into a lateral bend and twist to stay in line with the monitor. Participants had greater shoulder flexion/extension range of motion while performing the Front-to-Back motion while using the AC2 chair. Participants using the AC2 chair also had significantly more abduction/adduction and rotation in a Side-to-Side movement in the shoulders; the sitter was required to have both hands on the desk to stay stable while doing both motions (Side-to-Side and Front-to-Back) which could help explain the increase in shoulder movement. Performing both motions while seated in the AC2 chair pushes the body away from the mid-line of the computing position, this increased the range of motion in flexion/extension, lateral bend, and twist. When participants used the AC1 chair, the upper body remained stationary while the lower limbs moved to perform the pedalling and sliding forward movements, which would help explain the differences found between the AC1 and AC2 chairs. In a real-world work scenario, based on our head and shoulder posture findings, talking on the phone or taking a micro break away from the computer, would be an appropriate time to use the full range of motion of the active component of the AC2 to obtain the physiological benefits (i.e. increasing oxygenated blood in the gastrocnemius) (L� eger et al. 2022). The full range of motion of the AC2 from the current study compromised the recommended ergonomic computing position from the CCOHS guidelines (Canadian Centre for Occupational Health and Safety 2018), which could affect the level of attentional demand while computing and could lead to an increase in discomfort levels long term. The AC2 has several range of motion resistance settings: The highest setting on the AC2 can bring the chair to a near-static state, which is ideal for more cognitively challenging tasks. Treadmill desks have been reported to reduce mousing and typing speed compared to a static sitting posture because the whole body is in movement and required a greater attentional demand from the individuals (John et al. 2009). Doroff et al. (2019) also found that when participants sat on an unstable chair, it was significantly more difficult to read than when seated on a static chair. Doroff et al. (2019) found that participants had trouble reading because their whole body was in motion when the text was on the table, which is most likely what happened in this current study while using the AC2.
For the hips and the knees, both active chairs allowed the sitter to move in flexion/extension, lateral bend, and twist motion significantly more than the control chair, with no dominant significance between AC1 and AC2 chairs. The range of motion of the pelvis while seated in both active chairs was significantly greater in anterior-posterior tilt, lateral bend, and twist compared to the control chair. Between both active chairs, the Front-to-Back motion allowed the sitter on the AC2 to have more pelvis flexion/extension movement than the AC1. For the Side-to-Side motions, the AC2 had more pelvis lateral bend and twist movement than the AC1. For the trunk-pelvis angle, a greater range of motion in flexion/extension was found using both active chairs while moving Front-to-Back. During Side-to-Side movements, greater lateral movement and twists were found using both active chairs vs. a control chair. When the pelvis tilts anteriorly, the lumbar curvature will rotate into the recommended neutral lordosis (Le and Marras 2016). Both active chairs alternated between anterior and posterior tilt of the pelvis, co-contracting the trunk muscles would be a considerable strategy to maintain the lordotic curve (Watanabe et al. 2014) and decrease stress on the passive structure of the spine (Richardson, Toppenberg, and Jull 1990) caused by prolonged sitting. Lastly, the range of motion of the trunk-thigh angle in the AC1 was found significantly greater in flexion/extension compared to the other three workstations. The AC1 was designed to allow the sitter's lower limbs to slide forwards and open the trunk-thigh angle while maintaining the alignment of the head, shoulders, and arms when computing. A more open trunk-thigh angle up to �135 � has been found to help maintain the neutral lordosis curve to avoid discomfort and loading on spinal structures (Fleischer, Rademacher, and Windberg 1987).

Participant's perception of design and discomfort
Participants associated standing, with a significantly higher-level of lower leg discomfort compared to all three chairs, on the Rate of Perceived Discomfort Questionnaire. The AC1 was rated with the least low back discomfort, which aligns with the purpose of the AC1 design by opening the trunk-thigh angle to relieve spine compression, yet, it was not significantly lower than the other workstations. However, the participants rated the buttock area on the AC1 with a significantly higher level of discomfort compared to the AC2, the control chair, and the desk. Reported discomfort on the buttocks started after 30-min of sitting in the AC1. The AC1 chair scored low in the low back discomfort but high in buttocks discomfort; this would suggest that the active component of the chair may not be the cause of the buttocks discomfort but rather due to lack of seat pan support around the buttocks. Participants in the Cardoso, Cardenas, and Albert (2021) study reported less buttocks discomfort when using the original AC1 chair (just a split-seat pan design) over the control chair. As a result, the buttocks discomfort, observed in the current study, is most likely not caused by the split seat pan feature but due to the seat pan contour itself. For the current AC1 chair, some seat pan contour adjustment is recommended.
With the end of the session survey, participants reported that the control chair provided significantly fewer physical symptoms compared to the AC1 and the desk. According to the RPD questionnaire, the AC1 provoked more buttocks discomfort and the desk provoked greater lower back and legs discomfort. Despite the fact that participants rated the control chair more comfortable, both active chairs had substantially more biomechanical benefits than the control chair, including increases in gastroceminus neuromuscular activity, flexibility to open the trunk-thigh angle, and pelvis anterior tilt. All of the biomechanical changes found in this current study, contribute to a reduction in perceived discomfort caused by spine unloading tension over time. If participants worked at each of the workstations for longer periods of time, it is hypothesised that the active workstations would outperform the control chair in reported perceived discomfort.

Limitations
Given that the AC1 is a prototype it lacked adjustability and therefore some modifications are recommended. For participants taller than 6 feet, the AC1 could not be raised to the recommended height required to achieve a knee angle of more than 90 � . A lower seat pan prevented the tall sitters from having a greater range of motion when actively pedalling or sliding forward. A second recommendation would be to modify the seat pan to incorporate a softer surface and a contour around the buttocks to reduce the reported discomfort. The contour of the seat pan should help offload and distribute the pressure under the ischial tuberosities on other body surfaces. A more distributed pressure throughout the seat pan has been proven to help reduce perceived discomfort during a short seated task (Wang et al. 2019). A shortened seat pan depth would permit smaller users to engage with the backrest while sitting properly in the chair and perform both active tasks.
The lack of significant differences found in the active chairs compared to the control chair for the external obliques muscles suggests that proper use of both active chairs requires some training. Individuals should learn how to: 1-engage the deep core muscles before performing any of the motions provided by both chairs to minimise the possibility of causing pressure on the iliolumbar ligaments (possibly causing pain receptors in the sacroiliac region to fire) (Cardoso, Cardenas, and Albert 2021); and 2-make good use of the backrest in the AC1 while they are not peddling or sliding forward, allowing the superficial trunk muscles to rest (Pynt 2015).
The increased neuromuscular activity in the external obliques while standing could be attributed to the cold temperature in the laboratory. The low temperature prevented our participants from sweating during the trial and allowed the equipment to adhere to the skin. The participants were also wearing minimal clothing to ensure marker visibility, which increased their chances of feeling cold and contracting their obliques.
The first language of the majority of the participants was French yet all the text they needed to transcribe and all of the questionnaires were in English. It is possible that having the text in English might have negatively affected the level of productivity or typing speed for the typing task because they were not used to type in English (even though this was not measured); this could, however, potentially help explain the increase in head flexion while typing.
Gender, age, handedness, and anthropometry comparisons were not made in this specific study: this will be the subject of future research communication that will include physiological and biomechanical distinctions.
Only 1 h at each workstation was investigated during our experimental study. Studying the long-term effects of using an active chair vs. static chair would be advantageous to a workplace environment.

Conclusion
The study provided insight into the biomechanical changes of sitting in an active chair. Overall, when actively sitting in both active chairs, positive biomechanical effects were found when compared to the control chair and a standing workstation, which was also found in our phase 2 study (Cardenas et al. 2022). While using the active chairs during the web-browsing task (participants had to follow the active protocol) participants had the ability to move their lower limbs and trunk while seated without inducing high neuromuscular activity, implying that no physical fatigue occurred while moving (see L� eger et al. 2022 to further support this finding). However, when giving the choice of actively sitting or not during the typing task, individuals tend not to utilise the active components of the active chairs, which was also seen in phase 1 (Cardoso, Cardenas, and Albert 2021). To rectify the lack of use of the active components of the active chairs, multiple training sessions are recommended to help develop long-term healthy behaviour habits at the workplace.